Posteverywhere MCP

What posteverywhere/mcp MCP server does

The posteverywhere/mcp MCP server exposes PostEverywhere’s social publishing workflow to MCP clients. It can work with connected accounts on Instagram, LinkedIn, TikTok, X, YouTube, Facebook, Threads, Pinterest, Bluesky, Discord, and Telegram, as listed by the project description.

Agents can inspect account health and account-level details, including token expiry and whether an account can publish. Post tools support immediate publication, future scheduling, draft creation, edits, deletion, status filtering, and retrieval of per-platform results. Publishing results can include destination-specific errors and live URLs.

Drafts provide an approval workflow. An agent can save content without publishing it, retrieve drafts and their platform-specific content, and later schedule or publish an approved draft. This is useful when generated content needs review before it reaches any social account.

How it works

The package communicates with an MCP-compatible client through a standard local stdio process. The client starts the npm package, and the process uses the POSTEVERYWHERE_API_KEY environment variable to authenticate with PostEverywhere. The README also describes a hosted connection option, but the package setup shown for Claude Desktop and Cursor uses the local npm process.

A typical workflow begins by listing connected accounts, then creating either a post or a draft. Scheduled and draft posts can be updated or deleted. If publication fails for some destinations, the agent can inspect the individual results and retry failed destinations. Platform filtering and status filters help an agent find scheduled, published, failed, or draft content.

Media operations use the PostEverywhere library. Existing files can be listed, inspected, or removed. A public image URL can be imported in one call; the returned media identifier can then be supplied when creating a post. The image-generation tool accepts a text prompt and supports four models and seven aspect ratios.

Setup and configuration

Create a PostEverywhere account, connect the social accounts to use, and create an API key from the developer settings. The README identifies read, write, and AI scopes: read access covers accounts, posts, and media; write access covers post and media changes; and AI access covers image generation.

For a local installation, configure an MCP client to run:

npx -y @posteverywhere/mcp

Set POSTEVERYWHERE_API_KEY in the client environment. Claude Desktop and Cursor examples use the package above with the API key in an env block. The project also states that Claude Code, Claude Desktop, ChatGPT through the hosted connector, Cursor, OpenAI Codex, and other MCP-compatible clients can use the service, although this directory entry provides client-specific configuration for Claude Desktop and Cursor.

Tools and capabilities

The listed tools include:

• Account listing and account health inspection.
• Post creation, scheduling, approval, updating, deletion, and retrieval.
• Status- and platform-filtered post listing.
• Per-destination publication results and failed-destination retries.
• Media library listing, inspection, URL import, and deletion.
• AI image generation with prompt, model, and aspect-ratio options.

Limitations and notes

The posteverywhere/mcp MCP server requires a PostEverywhere API key and connected social accounts; the MCP package does not itself establish those accounts. Platform availability and publishing status are exposed through PostEverywhere, so an account may report that it cannot currently publish or that its token is nearing expiry.

The supplied setup documents a local stdio server rather than a self-contained social network connector. Drafts remain unpublished until a later scheduling or publish action. The README points to separate API documentation for complete tool parameters and to a separate Node SDK for programmatic integrations.